Experiments (notebooks/)

This directory holds the artifacts for creating, fine-tuning, and managing the initial experiments for MobileViT-S using the SIPaKMeD dataset.

Training the Classifier (train_mobilevit_s_on_SIPKAMED.ipynb)

Allows execution and logging for all training runs directly in a browser environment (Jupyter). This notebook documents: - Splitting raw datasets. - Tuning hyperparameters for learning rates. - Generating epoch accuracy graphs. - Evaluating the model on the test dataset. - Saving weights out to .pth format for production.